01:03 <FromGitter> <ezrast> So this is weird https://play.crystal-lang.org/#/r/1kum

01:04 <FromGitter> <ezrast> I can't believe that's a minimal repro, but take away any of the code that shouldn't do anything and the error stops happening.

01:22 <FromGitter> <MaxLap> Wow, i love that the puts needs to have at least 2 characters like that for the crash to occur. This has to be a compiler bug.

01:30 <FromGitter> <ezrast> Oh good, I'm not crazy

03:26 <travis-ci> crystal-lang/crystal#4d03a56 (master - Fixed wrong username in changelog (#3929)): The build passed. https://travis-ci.org/crystal-lang/crystal/builds/194132029

03:26 <DeBot> https://github.com/crystal-lang/crystal/pull/3929 (Fixed wrong username)

15:00 <crystal-gh> [crystal] bcardiff pushed 1 new commit to master: https://git.io/vM59X

15:00 <crystal-gh> crystal/master 15032cc Brian J. Cardiff: Playground: fix .cr files served in workbooks. end of lines were chomped

15:43 <travis-ci> crystal-lang/crystal#15032cc (master - Playground: fix .cr files served in workbooks. end of lines were chomped): The build passed. https://travis-ci.org/crystal-lang/crystal/builds/194495285

19:36 <FromGitter> <drosehn> Hmm. Did the handling of NaN get sorted out? NaN is defined to be not-equal NaN, and that is not a weird decision. `0.0/0.0 == NaN`, but so does `42*0.0/0.0`. But you don't want the language to claim that `0.0/0.0 == 42*0.0/0.0`.

19:37 <FromGitter> <drosehn> NaN is *not* some specific value. It's a place-holder to say "whatever this value is, the computer has no way to represent it".

19:38 <Papierkorb> > But you don't want the language to claim that `0.0/0.0 == 42*0.0/0.0`.

19:38 <Papierkorb> Why, I mean, that's a correct statement math-wise

19:38 <FromGitter> <drosehn> Imagine NaRN, for "Not a Rational Number". pi and e are both "not a rational number", but that does not mean `pi == e`.

19:39 <Papierkorb> but `pi.narn? == e.narn?` would hold true

19:40 <Papierkorb> I get your point I think, but I also don't think there's a good solution to it

19:40 <Papierkorb> NaN is already pretty weird in math terms

19:41 <Papierkorb> A number that doesn't exist? huh?

19:41 <FromGitter> <drosehn> But it has to have the behavior that it has. In that sense, it is not weird. It's the only way to make it work right.

19:42 <Papierkorb> yes

19:44 <FromGitter> <drosehn> I'm not completely sure I understand where the earlier discussion left off. I just wanted to point out that you really don't want one NaN to be equal to some other NaN. Not unless the language keeps track of what generated each NaN.

19:47 <FromGitter> <MaxLap> You can see the discussion in the issues: https://github.com/crystal-lang/crystal/issues/3923

19:48 <FromGitter> <MaxLap> What you need is a different equality method for hashes (ruby has #eql? for this exact purpose)

19:49 <Papierkorb> Sounds reasonable

19:56 <FromGitter> <drosehn> There are a few cases where this causes a problem, and offhand I can only think of one. Note that in ruby: `bb = Hash.new ; bb[0.0/0] = 1 ; bb[0.0/0*40] = 2` gives you a hash which is: `{NaN=>1, NaN=>2}`

19:57 <RX14> adding more equality methods seems stupid personallt

19:57 <FromGitter> <drosehn> and `bb.has_key?(0.0/0) => false`

20:05 <Papierkorb> I don't have much stake in it tbh, but is it an actual problem? Adding an eql? equivalent has a cost, but is it worth it?

20:15 <Yxhuvud> papierkorb: no, mathwise it is an *undefined* statement.

20:15 <Papierkorb> Yxhuvud: are they different undefineds too?

20:16 <Papierkorb> (I guess so, but...)

20:16 <FromGitter> <mverzilli> @drosehn I think I agree with you

20:16 <Yxhuvud> statement doesn't make sense. undefined is undefined - there are no equality operation defined for it.

20:17 <FromGitter> <mverzilli> @papierkorb think of algebra of limits in Math

20:17 <FromGitter> <mverzilli> you can't claim inf == inf unless you know they "progress at the same speed"

20:22 <Yxhuvud> I kinda dislike that Hash[] is dependent on ==, but tbh, I don't feel it is worth caring too much about it. Any performance and most clean code concerns would probably be more important. It is an edge case that isn't so cute, but still an edge case.

21:35 <FromGitter> <drosehn> Hmm. I'm installing the newest crystal using `brew` on MacOS, and it tells me that "autocode is not installed". And `brew` does not know about any package called `autocode`. Que pasa aqui? (<- spanish from my high school days, many many years ago. Possibly incorrect spanish.)

21:41 <FromGitter> <drosehn> MacOS Yosemite (10.10.5), if that matters.

21:52 <FromGitter> <mverzilli> that's good Spanish :)

21:54 <FromGitter> <mverzilli> I'm on Sierra, but let me give it a try

21:56 <FromGitter> <drosehn> There are only a few phrases that I remember, but we used to say that one a lot.

22:00 <FromGitter> <mverzilli> looks to work alright on Sierra, I was going to suggest that you opened an issue, but I see you just did :)

22:01 <FromGitter> <mverzilli> just in case, could you run `brew doctor`and paste the results here or an issue?

22:02 <FromGitter> <drosehn> Well, actually I mentioned it in reply to a different issue.

22:07 <FromGitter> <drosehn> `brew doctor` complains: ```cr

22:09 <FromGitter> <drosehn> ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=58867f19e836bf7010817c41]

22:10 <FromGitter> <drosehn> ... it's in /opt/homebrew, because I didn't want `brew` to be doing anything in /usr/local (where I have some other packages).

22:11 <FromGitter> <drosehn> I don't seem to have any file named `autocode\*` under /opt/homebrew.

22:15 <FromGitter> <drosehn> There is a: ⏎ /opt/homebrew/Library/Taps/homebrew/homebrew-core/Formula/autocode.rb ⏎ I'll see if I can do something with that.

22:23 <FromGitter> <drosehn> Hmm. I think I've done something which may have fixed things. Or it may have broken them in different ways.

22:42 <FromGitter> <mverzilli> that's good, or bad

22:42 <FromGitter> <mverzilli> would you elaborate?

22:43 <FromGitter> <drosehn> Ugh. Well, I think I'm in a mess which is mostly because brew needs to do things in a way that I don't want it to. Namely, it wants to own /usr/local, and I want to box it into /opt/homebrew. I now have autocode installed okay, but now when I go to install crystal I'm told "autocode 1.3.1 already installed". Why, *yes*, it's installed. You said you needed it installed, and I had brew install it.

22:45 <FromGitter> <drosehn> This has me unhappy, but I don't want to rant about it here.

22:46 <FromGitter> <mverzilli> sorry about that! I'm not sure what's the right place to rant about this issue :worried:

22:54 <FromGitter> <bcardiff> @drosehn i think latests versions of brew are able to work without owning /usr/local. Also you can download crystal binaries from github if you prefer other installation methods

22:56 <FromGitter> <drosehn> Well, I don't mean "owning" as in the unix sense of `chown`. I mean they want to have total say in what goes on in that directory and all subdirectories. And I don't want `brew` to have *anything* to do with /usr/local.

22:57 <FromGitter> <drosehn> I may have a work-around though. I'll keep experimenting for awhile.

22:59 <FromGitter> <drosehn> I recently used `brew` to upgrade to a new version of `ripgrep`, and ran into problems there too. So I suspect I'm tripping over issues in the newer versions of `brew`, or that the fact that I'm still on yosemite is causing more and more problems as time goes on. Probably both!

23:15 <FromGitter> <drosehn> I might have to wait until tomorrow to continue experiments. We have a snowstorm coming in very soon, and I should head home! :smile:

23:35 <FromGitter> <crisward> Anyone can give me some help on json builder?

23:36 <FromGitter> <drosehn> I'm not sure I can help, but what's the problem?

23:37 <FromGitter> <crisward> ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=588693ca074f7be7630b6675]

23:38 <FromGitter> <crisward> This outputs json from a model with a json mapping. However, I'm not getting any commas after each row of data.

23:52 <FromGitter> <crisward> ok worked it out, thanks.

23:52 <FromGitter> <drosehn> I haven't worked at that level of json. You want a comma after each `model.new(resultSet).to_json(ctx.response)`? I'd be tempted to just add that as a separate command in the innermost `do` loop. I'll guess you'd do that with `json.raw(".")`

23:52 <FromGitter> <crisward> `model.new(resultSet).to_json(ctx.response)` should have been `model.new(resultSet).to_json(json)`

23:53 <FromGitter> <drosehn> Hmm. did that come out as a comma or a period? I meant to type a comma.

23:53 <FromGitter> <drosehn> ah.